[xml] Re: xmlIndentTreeOutput = 1??

["Thomas Rokohl" <rokohl raygina de>]

ok, i understand the problem so i try do make a workaround in my project.
i use more than one xml file and in one of them i will use 
xmlKeepBlanksDefault(0) and in
the other xmlKeepBlanksDefault(1).

so each function which use libxml looks now like this:

void functionA() {
int old = xmlKeepBlanksDefault(0);
...
xmlKeepBlanksDefault(old);
}

my project create the xmltree by it self if the file not exist.
and now i have the effect that if the file will be create there is no 
indent.
if i read the file a second time an write in again -> it have indent?!

i also try to make

xmlKeepBlanksDefault(0)
xmlSaveFormatFileEnc(filename.c_str(),doc,"UTF-8", 1);
xmlKeepBlanksDefault(1)

but it not work.

so my question is: to which functions xmlKeepBlanksDefault have affection ??


"Daniel Veillard" <veillard redhat com> schrieb im Newsbeitrag 
news:20041101213256 GI20274 redhat com   
> On Mon, Nov 01, 2004 at 07:16:22PM +0100, Thomas Rokohl wrote:
> > hi,
> >
> > i have a problem with the xmlSaveFormatFileEnc function.
> >
> > if i write:
> >
> > xmlIndentTreeOutput = 1;
> > xmlSaveFormatFileEnc("test.xml",doc,"UTF-8", 1);
> >
> > there is no indent in the file, but if i use:
> >
> > xmlKeepBlanksDefault(0)
> >
> > it works. the problem is that i don't want to set 
> > xmlKeepBlanksDefault(0).
> >
> > so what's wrong by the first way?
> >
> > i think the difference between this two way's for setting
> > xmlIndentTreeOutput is that den second way is an function call in the
> > library. but
>
>  The core of the issue is the following:
>    - if libxml2 sees a text node child for a node, it won't indent
>      the output of that node to avoid destroying mixed content data
>    - this policy also affects existing text nodes used for previous
>      indenting.
>
> xmlKeepBlanksDefault(0) removes all existing indentation nodes. Yes this
> is bad, it's an heuristic. But expecting xmlSaveFormatFileEnc() to add
> subch blanks or modify them is also an heuristic, as dangerous.
> In a nutshell only the application can really know if spaces are ignorable
> or not in an instance, and relying on the parser/toolkit to do so is not
> safe. There is no easy solution.
>
> Daniel
>
> -- 
> Daniel Veillard      | Red Hat Desktop team http://redhat.com/
> veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
> http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/